markjgsmith

Abi Noda of DX [56:35]: “How do you upscale developers? How do you optimise your code for LLMs? How should platform engineering teams think about self service and enablement? [...] AI tooling, AI development workflows, curating Claude workflows [...] how do you create a standardised set of workflows so when you hire a new developer in an organisation, boom, they have this menu of super powers [...] those types of adjacent problems are new constraints for enterprises looking to deploy AI at organisational scale. Not single player mode, but more multi-player mode. How does an ORGANISATION become successful with these tools? It's a different set of problems.” #

Several quite grueling days of heavy programming in a row, but I am starting to get somewhere with my devcontainer setup and with how I‘m organising my work with Gemini. Hoping to get some more time tomorrow to write something in more detail, but essentially it involves writing task specifications and tracking them in git. It‘s using the very cool Backlog.md project. The cool thing is that Gemini can generate the backlog commands as you move through your tasks, and so it‘s mostly trivial to totally keep on top of your progress, even if you get disconnected and have to start with a fresh Gemini connection, which happens very often. Anyway I think this could make a big difference to velocity and accuracy. #